Human Translation Soon to Become a Luxury Service
Written by Konstantin Josseliani
A modern translation firm is more a technology or production company than a classic team of linguists. We have an engineering department, an ERP system, a quality management system, as well as a number of technological solutions, e.g., the Global Technology Platform (GTP) we developed for taking orders and managing projects. Finally, like all leading translation providers, we actively use software to generate translation memories, as well as professional machine translation systems with subsequent post- editing. Many believe that some emerging technology will sooner or later kill our industry, but so far, the situation has been quite the opposite — professional use of this technology makes us more competitive and allows us to translate large volumes of content.
The translation business in our country is extremely fragmented — thousands of small companies operate on the market. The Russian portal Translation Rating estimates total sales at Russian translation companies to be 17B rubles. Last year, Janus Worldwide was rated 45th on the Top 100 Language Service Providers around the world compiled by Common Sense Advisory. Another Russian company — ABBYY LS — was rated 50th. We receive around 85% of our profits from international customers, this year we will grow 20-25% compared to 2016. Our goal is to become one of the Top 30 global translation companies, and I hope technology can help us achieve that.
Technical Assignment

I graduated from the Moris Tores Foreign Language Institute in 1994. My first foreign language is German, and English is my second. Immediately after graduating, I got a job with SAP, which was a young company just entering the market. They needed software and technical documentation localized into Russian. Software localization has its peculiarities — the user sees commands and system messages on the screen out of context. Back then, SAP was already actively using computer-aided translation technologies. For instance, translation memory, which is a kind of repository for translations.
When the machine matches previously-translated elements of text and then automatically propagates them into subsequent documents, it speeds up the process significantly.
Two years later, I was the head of the localization department and managing 10 people. But I wanted to try something new. For instance, open my own business to provide services to IT companies – and the niche was practically empty. By that time, I had obtained another degree in global economics from the Financial Academy of the Government of the Russian Federation and then opened Janus with a friend from school. We worked with the premium end of the market — large Western companies. Naturally, SAP became our first client. Then we started working with HP, Cisco, Canon, Epson, ZyXEL and others. Word of mouth helped us, and Janus grew to employ 15 people within a year. Three or four years later, we decided to scale up the business and picked new directions.
We started with foreign organizations that didn’t have offices in Russia but wanted to sell their goods and services on the Russian market. Most of them are IT companies or manufacturers of technically complex products like equipment and cars. The second direction was Russian companies. We started working with Russian Railways, Siemens, ABB, and other large customers. Here is where Janus first encountered serious competition.
Adding Offices

In order to set ourselves apart from other players, we offered customers translations not only into Russian but also into the so-called CIS languages, especially more popular Ukrainian and Kazakh. We were the first ones to open offices in Kiev and Almaty. It was a serious investment, because we hired and trained people (translators, editors, project managers) first and then started looking for customers.
But this immediately let us get ahead of our competitors — we were first considered to be a single-language vendor (SLV), i.e., a company that provided translation services into one language, but after opening new offices, we became a regional language vendor (RLV).
It is easier for the customer to order a translation from the same company than looking for contractor in different countries.
After CIS countries, we started opening additional offices in Russia. In the translation business, payroll accounts for about 30% of expenses, which is why opening offices in regions helped us save a lot of money. We picked cities with universities that provided good linguistic education: first, Nizhnii Novgorod, then St. Petersburg and Tomsk. Today, we have 50-100 people working in each regional office.
Even though we cut costs, it was difficult to grow any further. The large companies with which we worked followed corporate rules and could not give all their orders to one contractor. We already occupied a significant portion of the technical and IT translation market, and further expansion in Russia would have required too much effort. We had essentially reached our limit. Our only choice was to enter the global market and become a multi-language vendor (MLV), i.e., provide companies with translations into all languages. In 2012, we developed a globalization strategy. The real problem was the time zones. A supplier working with global corporations must accept orders at any time of day or night. For example, if a US-based company needs to translate a document from English into seven other languages, a project manager must respond immediately and not in six to seven hours.
We decided to apply the “follow-the-sun” concept, i.e., open offices in different countries in order to always stay in touch. Within several years, we’d opened offices in Germany, Britain, USA, China and the Czech Republic, and soon we will open one in Argentina. Each foreign office has its own payoff period, but the average term is 18 months. We get requests no matter what time it is. For some companies, we regularly translate texts into 60 languages.
Recently orders have gotten smaller, but more and more of them come in, and almost all of them are rush projects. In the past, translators could quite literally receive a whole filing cabinet of documents to translate, but now we deal with much smaller portions. I believe this is due to IT companies using agile technologies, which is why they only require small changes.
Our staff also had to adapt to this schedule. We employ only 300 employees, and 90 of them are in Moscow. Several years ago, we developed a flexible incentive system – in addition to salaries, we give bonuses to employees for excellent performance and quality. How we calculate bonuses is rather complex, but every person can see his/her indicators on their profile. Thanks to this incentive system, we’ve achieved significantly reduced employee turnover. People realized they could make more money if they made more effort. In addition to working with full-time employees, we outsource work to three thousand people in different countries. Resource Management is in charge of them – they find people and then test and train them, and we agree on the terms of their service.
Engineering Design

The market is lead by large companies like Lionbridge and TransPerfect that each account for about 10% of the global translation market. We can only compete with them using technologies, which is why the translation business and the IT business are growing together.
First, in order to localize software, a company needs engineering knowledge. Applications have to be tested to make sure they work properly when receiving commands in other languages. Second, everyone uses translation memory systems.
Third, many companies have recently started using machine translation. I want to say right away that I don’t mean Google Translate. I am talking about professional systems applied exclusively in our industry. For example, Asia Online and Systran, and Microsoft also have similar software, etc. Today, machine translation systems are extremely specialized – they are designed exclusively for technical texts written in a special language. This is so-called controlled language, in which sentences are built with a basic structure, basic enough for a preschooler to understand. The machine is loaded with terminology in a specific area, it provides a translation, and then a human corrects it and loads it back into the machine. As opposed to a human, the machine does not make factual errors, does not confuse numbers, and accurately translates terms. The worst it can do is to create an awkwardly- built sentence. That is why the industry is seeing increased demand today not for regular editors, but for post-editors. This idea is very popular in the West.
The rise of machine translation would invariably affect the rates charged by translation companies. Today, many customers send us texts which have already been pre-translated by a machine, and all we can do is post-edit them. They of course ask for discounts on services. Everything depends on a specific language, but in most cases, we’re talking about prices going down 25-30% compared to human translation. This is simply a fact we have to live with. But this has a positive side as well. Thanks to software, the amount of content that requires translation is growing exponentially. And thanks to this increase in volumes, the industry itself is growing. In a word, when it comes to software, documentation, and other reference materials, we have already lost the battle to machines. On the other hand, machines still can’t process texts that utilize the richness and figurativeness of language, such as books, marketing materials, press releases, and site content. Machines still don’t understand fixed expressions. So human translation will soon be considered a luxury service.
Translation Management
Translation companies also need technologies for project management. Businesses cannot be scaled up without them. For instance, we implemented 1C ERP, which we modified to our needs. As a result, all processes like sales and marketing, etc., have been automated. In addition, we developed our Global Technology Platform for working with customers and translators. The platform includes several modules.
The customer service module formalizes our interactions with the customer. An order can be placed via the customer’s account, and our managers will receive an automatic alert, put together a quote, and once the customer approves it, the text is then sent out for translation. The customer can thereafter monitor the status of its project. Moreover, the customer can view all of the statistics regarding its interaction with Janus Worldwide, who ordered the translation, for what amount, etc.
The translator interaction module is built on the same principle. Translators are provided with a project notification, and if s/he is ready to take on the text, s/he confirms this by clicking on a link. It’s just like how Uber or other similar services’ drivers accept fares. The project manager then provides the translator with access to the text, and after the work gets done the files are sent back.
We developed Janus Express to handle rush translation demand. A number of intermediate steps have been taken out of the equation. For example, the order is charged automatically and sent to translators without project managers stepping in. We only allow trusted translators to take on projects from Janus Express who have earned a high internal rating. In the end, a job can be turned around in a few hours. At this speed, the translation review or editing stage is eliminated, and nor can we engage in quality control. But we accept this risk knowingly. In so far as experienced staff are involved, we’re confident quality will be acceptable. We’ve seen a lot of demand for this service. It’s fast, and it’s also clear why: there’s a minimum of human involvement.
We designed Janus Perfect to be our quality control system. Translated texts may be uploaded into Janus Perfect, and it will return a detailed evaluation. Editors evaluate translation quality according to predefined criteria, e.g., semantic accuracy, terminological accuracy, formatting, etc. This is another useful and in-demand service. Finally, customers aren’t just looking for quick translations but also to have convenient communications with their translation provider. Many customers have their content on internal computer systems, which wastes time sending it here and there. So we decided to develop another useful feature, Janus Connect. It can be connected to the customer’s computer systems, upload content for translation and then download translations back into the customer’s system.
It’s very unlikely the translation business will disappear in the foreseeable future, even if most of Russia happens to learn foreign languages (which is improbable). I see companies that are performing technical translation under less of a threat. Perhaps people will learn English to engage in everyday discussion or to listen to a foreign speaker at a conference. But sometimes the discussion is rather about documentation on electrical power stations, which is important because workers will soon be operating it. Or medical staff will be using a medical device. In each event the end user should be given information in his/her native language. So I think there will continue to be demand for our services for some time to come.